Design a scalable consumer feature

Q: Design a scalable consumer feature

This question evaluates a candidate's ability to design large-scale recommendation systems, covering system architecture, real-time serving and offline pipelines, personalization and ML modeling (candidate generation and ranking), data modeling/feature stores, caching and storage, monitoring, experimentation, and privacy/compliance skills.

Q: How do I approach System Design interview questions?

System Design questions require understanding of core concepts and practice. PracHub provides solutions with explanations to help you master system design interviews.

Q: What difficulty level is this interview question?

This is a hard difficulty System Design question, commonly asked during Onsite rounds at Airbnb.

Q: What role is this question designed for?

This question is commonly asked for Software Engineer candidates at Airbnb during technical interviews.

Question

System Design: Scalable, Privacy-Conscious Recommendations Service for a Consumer App

Context

You are designing a real-time recommendations service for a global consumer app (e.g., marketplace-style content such as listings or experiences). Assume:

Scale: 30–50M MAU, 5–10M items, traffic across Americas, EU, and APAC.
Surfaces: Home feed (top-N), in-search suggestions (inline), item detail page (similar items), emails/notifications.
Privacy: Operates under GDPR/CCPA and regional data residency constraints.

Design the service end-to-end, covering online serving and offline pipelines, while balancing relevance, latency, and privacy.

Requirements to Cover

Functional requirements

What the service must do, supported surfaces, personalization scope, and fallbacks/degradation behavior.

APIs

Online serving API for recommendations (request/response contracts).
Feedback ingestion APIs (impressions, clicks, conversions, hides).
Consent/identity APIs as needed.

Data models

Core entities: user, item, and interaction events.
Feature store schema (online/offline), embeddings, and versioning.

Personalization approach

Candidate generation and ranking stages.
Modeling choices (e.g., collaborative filtering, two-tower retrieval, session-based models).

Cold-start strategy

New user and new item strategies, exploration, and content-based fallbacks.

Ranking and feedback loops

Multi-objective ranking (e.g., CTR, conversion, value), calibration, diversity/novelty.
How to log and learn from feedback; avoid bias and clickbait.

Latency/SLA targets

SLOs/SLIs and a latency budget across components, with degrade modes.

Caching and storage choices

Online stores (vectors, features), caches (edge and per-user), invalidation/TTL.

Backfill and reprocessing

Event-time processing, replays, feature backfills, and reproducibility.

Monitoring and alerting

System health, data quality, model performance, drift, and runbooks.

Experiments support

A/B testing, holdouts, sequential testing/bandits, and guardrails.

Abuse/spam prevention

Shilling, bot/fraud detection, adversarial content.

Regional privacy/consent handling

Consent gating, data minimization, data residency, erasure, audit.

Make reasonable assumptions where needed and call them out explicitly.

Design a scalable consumer feature

Quick Overview

Design a scalable consumer feature

System Design: Scalable, Privacy-Conscious Recommendations Service for a Consumer App

Context

Requirements to Cover

Submit Your Answer to Earn 20XP